00:07
<paul_irish>
z-index, of course requires an element to be positioned, but positioned elements often don't play well with transformed elements. but i want to manage z-order of my transformed elements. is it worth considering z-index apply to tranformed elements or should vendors address issues of positioned transformed elems?
01:28
<Hixie>
man, can you imagine how the last 10 years on the web would have been if html4 had defined the semantics of elements to the level of care and detail that y'all are making me put into the elements' definitions now?
01:28
Hixie
updates the wording of the definition of <article> once again
01:31
<TabAtkins>
Hixie: Good or bad?
01:32
<TabAtkins>
paul_irish_: How does z-index and transformations play badly together?
01:32
<micheil>
Hixie: you could put in a claus that every time the spec gets updated for a minor word change that a kitten gets killed.
01:32
<Hixie>
TabAtkins: good, i hope
01:33
<Hixie>
micheil: that would be bad given how many minor mistakes i make
01:33
<aho>
padding + transform is really weird imo
01:33
<micheil>
heh, it was more a reference to yesterday
01:33
<micheil>
http://drp.ly/1rhtfH
01:33
<aho>
eg 0 10px + 90° = 10px 10px ish
01:34
<aho>
(is it really supposed to be like this?) :>
01:42
<TabAtkins>
aho: Yes. transforms don't alter the geometry of the object - it still takes up exactly as much space as it did before. So, the space previously taken up by the horizontal padding still exists, but the element also appears to now have "vertical" padding.
01:48
<myakura>
looks like http://www.whatwg.org/specs/web-apps/current-work/complete/ is gone. something gone wrong? politics?
01:50
<TabAtkins>
It's not gone on purpose.
02:25
<Hixie>
myakura: fixed
03:34
<myakura>
Hixie: Thanks!
07:54
<hsivonen>
what makes Web authors declare iso-10646 in meta?
07:55
<hsivonen>
...on pages that aren't in UTF-16BE
07:58
<Rik`>
copy paste ?
08:04
<hsivonen>
where would you copy iso-10646 from?
08:05
<estellevw>
lol
08:06
<Matjas>
How would you guys go about marking up a FAQ? I’m thinking of using a <dl> but I’m not sure if that's a good idea.
08:06
<Matjas>
i.e. <dl> <dt>Q: foo</dt> <dd>A: bar</dd> </dl>
08:06
<hsivonen>
Matjas: that works
08:07
<estellevw>
well, is the answer really defining the question?
08:07
<Rik`>
hsivonen: maybe wrong interpretation of http://www.w3.org/TR/html4/charset.html ?
08:08
<estellevw>
I think <dl> is very oftne, if not generally used for FAQ, but that's not quite semantically correct
08:09
<Hixie>
estellevw: it is in HTML5 as far as I can tell
08:09
<Hixie>
indeed the example for <dt> is a FAQ: http://www.whatwg.org/specs/web-apps/current-work/complete.html#the-dt-element
08:15
<estellevw>
i was looking here: http://dev.w3.org/html5/markup/dt.html, name or term
08:16
<Hixie>
that's not the html5 spec
08:16
<estellevw>
yeah, i know
08:16
<estellevw>
thinking that page needs to be updated to reflect that ther ehas been a change
08:17
<estellevw>
that is the quick cheat sheet that a lot of implmenetors are using
08:17
<Hixie>
HTML5 itself is pretty clear that "name" is not being used in the english sense here but in the sense of an association list
08:18
<Hixie>
but I guess it could be made even clearer
08:18
<Matjas>
Thanks hsivonen, Hixie and estellevw!
08:18
<annevk>
estellevw, which implementors?
08:19
<estellevw>
There was a lot of discussion around DL at openwebcamp last week
08:19
<Hixie>
i assume estellevw means authors, not implementors
08:19
<estellevw>
front end engineers
08:19
<estellevw>
yes
08:19
<estellevw>
sorry
08:19
<Hixie>
np :-)
08:19
<estellevw>
i'm a front end. I implement at work ;)
08:19
Hixie
has filed a bug to make HTML5 clearer about this btw
08:21
<estellevw>
the discussion revolved around the semantics, screen readers, accessibility, etc. I htink the conclusion was that it wasn't quite semantic, but it was the best solution. Glad to see the nuanced change in the HTML5. Good change. And, good to get an official clarification
08:21
<estellevw>
Plus, the next time I see everyone, i can say "I told you so" :D
08:27
hsivonen
believes the relationship of semantics and screen readers gets exaggerated
08:31
<estellevw>
hsivonen: Open Web Camp, had several visually impaired users in attendance. It made for interesting conversation, and blew away many notions of what people had assumed were tenets of accessibility.
08:31
<Hixie>
like what?
08:32
<estellevw>
Many believed the title attribute to be very useful for accessibility.
08:32
<Hixie>
current screen readers make it basically impossible to read the title="" attribute
08:32
<estellevw>
seems that they're not helpful at all for users of screen reader. y
08:32
<Hixie>
it's a major bug in current screen readers
08:32
<estellevw>
yeah
08:33
<Hixie>
seems so easy to fix, too, dunno why they don't
08:33
<estellevw>
probably because most authors abuse the title attribute for feeding google
08:33
<Hixie>
(just have an audio cue when a title="" is available on the element being read out, and have a key to immediately play back the titles of the text currently being played)
08:34
estellevw
trying to remember 'author' not 'implementor'
08:34
<Hixie>
(plus something to navigate by title, but that can be added to the many existing navigation mechanisms)
08:36
<Hixie>
user/reader, author/webdev/coder/designer/page/app, implementor/vendor/ua/browser
08:36
<Hixie>
the three main classes
08:36
<zdenekkostal>
Could you say in HTML5 spec. how screen readers should read and understand the content? I think it would be very usefull for web developers, because when I code some project, I haven't enought time to test it in screen readers...
08:37
<estellevw>
I don't test in screen readers either
08:38
<estellevw>
the best quick accessibility checks involve navigating your page without your mouse (just use the keyboard), and test the page without images and without css to see how the screen reader would read the page (leave javascript on)
08:38
<Hixie>
if you're on a mac you can use voiceover to test it
08:39
<estellevw>
Windows 7 has a very cool reader as well.
08:39
<estellevw>
unfortunately (or fortunately) I don't have windows 7 yet
11:32
<annevk>
so unless something unforeseen happens I'm going to Maastricht after all
11:41
<hsivonen>
annevk: how long is the meeting? I thought it would be over by now.
11:42
<annevk>
it's tomorrow midday
11:42
<annevk>
I'm just going for hybi
11:48
<hsivonen>
annevk: "fun"
11:49
<hsivonen>
so in the meta charset sniffing "UTF-16" is special-cased
11:49
<annevk>
we'll see
11:49
<hsivonen>
does that really mean "UTF-16" only or also BE, LE and 32?
11:49
<annevk>
hopefully talking F2F will help clarify a few things
11:49
<annevk>
UTF-32 is irrelevant
11:50
<annevk>
iirc it just means UTF-16 case-insensitive
11:51
<hsivonen>
ok. how about the other cases in the spec where the charset of a URL is checked for "UTF-16", does that mean BE and LE, too?
11:51
<hsivonen>
(my guess is yes)
11:51
<annevk>
oh sorry
11:51
<annevk>
it says "a UTF-16 encoding"
11:51
<annevk>
I guess that means all of them
11:52
<hsivonen>
aliases, too?
11:52
<hsivonen>
it seems to me Safari doesn't do aliases
11:53
<hsivonen>
UTF-16 is such a bad idea :-(
11:53
<hsivonen>
must have seemed great at the time
11:53
<annevk>
aliases too
11:53
<annevk>
in fact, lots of the encoding stuff are pretty bad ideas
11:54
<annevk>
one day i'll sort it out
11:54
<Ms2ger>
Good luck
11:54
<annevk>
and ICU can be obsoleted
11:54
<annevk>
I made quite a bit of progress with encoding detection
11:54
<hsivonen>
annevk: it's good to have goals
11:54
<annevk>
and encodings to be supported by popular user agents
11:55
<annevk>
what there's yet to do is defining for each encoding how it maps to Unicode
11:55
<annevk>
(and possibly how it maps back)
11:55
<hsivonen>
annevk: doesn't ftp.unicode.org already do that except for Microsoft deltas?
11:55
<annevk>
in more detail than encodings are defined today (e.g. precise error handling for UTF-8) and more compatible (lots of ASIAN encodings support more than the standards indicate)
11:55
<hsivonen>
annevk: ok.
11:57
<hsivonen>
Woohoo. at least this iso-10646 thing isn't my mistake but a spec bug :-)
11:57
hsivonen
files
12:18
<hsivonen>
Philip`: do you happen to have something that you could grep for pages that say iso-10646 in meta and don't have an HTTP-level charset?
12:19
<hsivonen>
I'd like to know if those pages work better as UTF-8 or as Windows-1252
12:19
<hsivonen>
or UTF-8 vs. chardet + default
12:22
<hsivonen>
Philip`: it would be great to have similar data for UTF-16BE, UTF-16LE, UTF-32, UTF-32BE and UTF-32LE
12:24
<Philip`>
hsivonen: http://philip.html5.org/data/charsets-2.html#charset-iso-10646-1
12:25
<Peter`>
Ericsson seems to be playing around with implementing <device> in WebKit: https://labs.ericsson.com/blog/beyond-html5-implementing-device-and-stream-management-webkit
12:25
<hsivonen>
Philip`: thanks! (looks like a rare value, thankfully)
12:25
<Philip`>
hsivonen: Also http://philip.html5.org/data/charsets.html#charset-iso-10646 from a smaller but possibly more reliable set of pages
12:27
Philip`
doesn't entirely trust his data analysis, particularly for non-ASCII-compatible encodings
12:31
<annevk>
Peter`, pretty cool
12:31
<annevk>
Peter`, though not sure why they use video_player rather than media
12:32
<hsivonen>
the UTF-32 cases no longer say UTF-32...
12:33
<jgraham>
othermaciej: Are you or abarth going to be at the hybi meting?
12:34
<othermaciej>
jgraham: definitely not me
12:34
<annevk>
hsivonen, in the spec? UTF-32 support was dropped
12:35
<hsivonen>
annevk: the pages listed as UTF-32 in Philip`'s data
12:35
<Peter`>
annevk: apparently they use a hack to circomvent security limitations during development
12:38
hsivonen
mumbles about x-imap4-modified-utf7 and Java's modified UTF-8
12:43
<hsivonen>
does webkit.org have a source code search that doesn't search changesets?
12:43
<hsivonen>
(I'm looking for the code that maps UTF-16 to UTF-8 in meta)
12:51
<Philip`>
hsivonen: Google Code Search with package:webkit sort of works, vaguely
12:51
<hsivonen>
dear lazy IRC, how do I make Ubuntu show me crash stacks for apps that have debug symbols but that don't come from Canonical?
12:51
<hsivonen>
Philip`: thanks
12:51
<Philip`>
Looks like it might be http://trac.webkit.org/browser/trunk/WebCore/loader/TextResourceDecoder.cpp?rev=64262#L344
12:52
<Philip`>
and http://trac.webkit.org/browser/trunk/WebCore/platform/text/TextEncoding.cpp#L224 which returns UTF-8
12:52
<Philip`>
perhaps
12:52
<jgraham>
othermaciej: OK. Do you have any opinion on the framing issue? Both technical and whether it is worth discussing in the "four week" timescale?
12:52
<hsivonen>
Philip`: interesting. thanks. that doesn't match my black-box testing, though
12:53
<othermaciej>
(I wonder if we should be saying "message" instead of "frame")
12:53
<hsivonen>
(or I misinterpret what "Default" means in Safari's menu)
12:53
<hsivonen>
does Default in Safari mean "what page said" or "user's default"?
12:53
<othermaciej>
jgraham: I think length-delimited text frames would be slightly technically superior to sentinel-delimited, but I don't care very much either way
12:54
<othermaciej>
jgraham: I think it's essential for text and binary to be clearly distinguished, and I think using presence/absence of a MIME type or whether the MIME type starts with text/ are very poor ways to signal text vs binary
12:55
<othermaciej>
jgraham: if we want to change text frames to be length-delimited, it probably needs to happen sooner rather than later
12:55
<jgraham>
othermaciej: OK, that mirrors my thoughts rather well. SO I am a little more confident that I am not toally off base :)
12:56
<jgraham>
I wonder if no one is discussing the handshake issue because they don't feel confident, or simply because the conversation has been led astray
12:57
<hsivonen>
Philip`: blame points to encoding, not decoding: https://bugs.webkit.org/show_bug.cgi?id=21635
12:58
hsivonen
wonders what proportion of Web users globally is behind a proxy
13:14
<hsivonen>
non-UTF-8 encodings lead to unhappiness
13:14
<annevk>
yeah
13:15
<annevk>
can't believe this Greg Wilkins is pushing for non-UTF-8 support in WebSocket
13:15
<annevk>
I often get the feeling with him he's just pushing for features
13:15
<annevk>
features and complexity
13:16
<hsivonen>
annevk: I trust Hixie is good at saying "no" in that case
13:17
<hsivonen>
annevk: what's the rationale for non-UTF-8?
13:18
<hsivonen>
Hixie: fwiw, when <acronym> is obsoleted, it would be logical to zap <strong>, <em>, <kbd>, <samp> and <cite>, too
13:19
<annevk>
UTF-16 was there and "potential future encodings"
13:19
<hsivonen>
(maybe not practical for validator user-friendliness but consistent with the removal of <acronym>)
13:19
<hsivonen>
annevk: that's so sad
13:21
<hsivonen>
If we ever run out of UTF-8, it's more likely to start allowing 6-byte sequences than to come up with something new
13:35
<annevk>
it in fact allowed those once
13:52
<annevk>
is there a counter proposal for http://lists.w3.org/Archives/Public/public-html/2010Jan/1006.html to keep the wiki in place?
13:52
<annevk>
having more things rely on IANA is a step backwards imo
13:53
<hsivonen>
what's the deal with having to log in to the W3C bugzilla all the time?
14:10
<othermaciej>
annevk: no one ever wrote a counter-proposal
14:18
<annevk>
meh
14:21
<crash\>
Must HTML5 browsers conform with the XHTML-Syntax and support application/xhtml+xml?
14:22
<annevk>
no
14:24
<crash\>
Is there anything about Processing Instructions in the descriped HTML parsing process?
14:24
<crash\>
are they simply skipped?
14:24
<annevk>
they are treated as bogus comments
14:25
<crash\>
ok, thanks
14:29
<Lachy>
othermaciej, given the new information about RFC 20 from Leif yesterday, how will that affect the current ASCII ref poll?
14:29
<annevk>
RFC 20 also does not refer to Unicode
14:44
<hsivonen>
the recent polyglot threads demonstrate why working on a polyglot draft is bad
14:44
<hsivonen>
it's a time sink for the WG
14:45
<Lachy>
it's especially bad when we start getting requests for polyglot-only syntax, which really goes against the whole concept of what a polyglot document is.
14:46
<hsivonen>
Lachy: indeed
14:57
<jgraham>
I really wish I had a clue why anyone wants to add MIME to WebSockets
14:57
<jgraham>
Or specifically Content-Type
15:17
<hsivonen>
argh. I got trolled into the polygot threads again
15:33
<Lachy>
does anyone know where the Mozilla's useragent stylesheet html.css file has gone? It no longer appears in the res directory, as it did in older Firefox builds? I'm trying to see exactly what the new styles for figure are in now in Minefield
15:34
<Lachy>
oh, same as blockquote.
15:35
<Lachy>
found it via DOM Inspector.
15:48
<Ms2ger>
Lachy, that's what the spec said
15:49
<annevk>
fun -- http://www.flumotion.com/first_webm_live_event.php -- WebM streaming
15:49
<annevk>
seems to work fine
15:49
<annevk>
although the the time and general controls seem a bit off
15:50
<hsivonen>
so weird to see "Oracle" on the projector
15:51
<hsivonen>
why do they only advertise Opera and Firefox? Is streaming broken in Chrome?
15:52
<annevk>
hmm, someone is talking about chrome usability, but it's not mpt afaik
15:52
<annevk>
hsivonen, maybe it's not in Chrome beta builds yet?
15:56
<annevk>
s/chrome/Ubuntu/ doh
15:56
<annevk>
well, GNOME
15:59
<zdenekkostal>
the WebM streaming is incredible! Only timeline in Opera seems broken
16:00
<hsivonen>
works great in Minefield. the timeline was a bit jumpy during the first minute but now it looks ok
16:05
<jgraham>
hsivonen: You know how you love document.write...
16:05
<Lachy>
doesn't work in chromium for me on Mac
16:05
<Lachy>
shows one frame, but doesn't play
16:05
<jgraham>
I just found some odd behaviour that seems to be a Gecko (+WebKit) bug...
16:16
<hsivonen>
jgraham: what's the behavior?
16:16
<jgraham>
hsivonen: One moment
16:19
<jgraham>
hsivonen: I can't reproduce over HTTP :(
16:19
<jgraham>
I think it is a race condition
16:19
<jgraham>
http://hoppipolla.co.uk/tests/document_write/001.html
16:20
<jgraham>
The odd behaviour in WebKit reproduces though
16:20
<jgraham>
hsivonen: When I run it locally I get output like:
16:20
<jgraham>
0,54,1,50,2,51,3,56,4,61,4,1280417173326
16:21
<jgraham>
It also didn't reproduce when I moved all the js inline
16:21
<jgraham>
(but did in WebKit, where I think it is not a race condition, just mildly crazy)
16:23
<hsivonen>
Is see 0,52,1,49,2,51,3,49,4,50 in Minefield over HTTP
16:23
<jgraham>
hsivonen: Yeah, that looks sane
16:23
<jgraham>
And is the same as I see over HTTP
16:24
<hsivonen>
ok. that *is* weird
16:24
<hsivonen>
jgraham: thanks
16:25
<jgraham>
hsivonen: (it is not a HTTP vs non-HTTP thing because it is wrong if I run it from a local apache)
16:25
<jgraham>
(so I assume it is speed)
16:27
<hsivonen>
speed of the external script load vs. the initial iframe load?
16:27
<jgraham>
Could be
16:28
<hsivonen>
maybe a local script gets read in as one event loop task
16:31
<hsivonen>
jgraham: filed as https://bugzilla.mozilla.org/show_bug.cgi?id=582975
16:33
<jgraham>
hsivonen: Thanks
16:36
<colapop>
jgraham: isn't start_time null at the time you set the event listener? Maybe the race condition is it fires & calls done() before start_time is initialized?
16:37
<colapop>
er, nevermind :)
17:08
<othermaciej>
I don't see where RFC 20 was mentioned
17:15
<TabAtkins>
othermaciej: In the latest message to the "Issue 101: us-ascii-ref - Straw Poll for Objections" thread.
17:15
<TabAtkins>
He linked http://www.ietf.org/rfc/rfc0020.txt
17:17
<TabAtkins>
Sigh. Okay, the fancy ascii table formatting is pretty, but really, what's the point of doing something where you have to split up the hexits when you can just do a simple two-column table? A split 2col table could be equally compact, or likely more compact actually.
17:17
<TabAtkins>
Silly people from 40 years ago.
17:18
<TabAtkins>
Or I guess 10 years ago, maybe? When it was put into the archives?
17:19
<othermaciej>
TabAtkins: was this on public-html? I don't see it...
17:19
<TabAtkins>
othermaciej: Yeah, it's the thread you started.
17:20
<jgraham>
othermaciej: http://www.w3.org/mid/20100728203158367745.c11b49f9⊙xn
17:20
<jgraham>
If the archive is working again
17:20
<TabAtkins>
What's the purpose of the /mid/ thing, when you can just dive into the individual directories for each list?
17:23
<othermaciej>
jgraham: hmm, oddly I don't see it in my public-html folder
17:23
<othermaciej>
checking if it might be somewhere else
17:23
<TabAtkins>
You been auto-spamming Leif's email?
17:23
<jgraham>
Maybe othermaciej has started snaking on email in the night
17:23
<jgraham>
*snacking
17:23
<othermaciej>
offhand, I am not sure what to do about it, unless someone seriously wants to propose it as an alternative instead of the ones already on the table
17:24
<TabAtkins>
Well, it's an official-looking and seemingly complete ASCII ref.
17:24
<TabAtkins>
And it was written by Vint Cerf!
17:25
<othermaciej>
hard to argue with that
17:26
<othermaciej>
is it in any way superior to the suggested ECMA-006?
17:26
<othermaciej>
my recollection from the poll is that the main objection to 006 is not defining the mapping to unicode, which flaw RFC20 seems to share
17:26
<TabAtkins>
More official-looking. Otherwise, no, because it lacks the unicode mapping.
17:27
<othermaciej>
I suspect it's actually less official
17:28
<TabAtkins>
Actual officialness is irrelevant.
17:28
<othermaciej>
ECMA-006 is the ECMA version of the joint publication with ISO that is the successor to ASCII
17:28
<othermaciej>
so it's an official joint publication, not just an ad-hoc copy
17:28
<othermaciej>
at least as I understand it
17:28
<annevk5>
yeah, it's just an RFC
17:29
<annevk5>
though really this is all quite nutty
17:29
<annevk5>
and quite the time sink
17:29
<TabAtkins>
Man, Cerf really liked his joke RFCs.
17:29
<othermaciej>
if we just let the poll continue then mainly it will consume the chairs' time and not much more of anyone else's
17:30
<annevk5>
the number of times this issue has been discussed here on IRC and on the list...
17:30
<annevk5>
anyway, back to super mario
17:33
<jgraham>
annevk5: Mario Galaxy 2?
17:36
<annevk5>
bros
17:37
<jgraham>
Ah
17:37
<TabAtkins>
kickin' it old school
17:37
jgraham
is wondering how desperate he should be for MG2
17:37
<annevk5>
bros wii that is
17:37
<annevk5>
mg2 is quite cool
17:37
<annevk5>
but some frustrating levels made me give up for now
17:38
<gsnedders>
jgraham: Aww, want it for your birthday?
17:38
<jgraham>
Mario Galaxy + Yoshi sounds like it has to be awesome
17:38
<jgraham>
(although most things + Yoshi sound like that)
18:26
<hsivonen>
so now that IANA has a link registry, it can't be used yet, because more bureaucracy is needed to establish certain flags in the registry???
18:26
<annevk5>
of course
18:26
<annevk5>
that's why we should have a wiki instead
18:34
<Workshiva>
Is the new registry properly registered?
18:35
<annevk5>
think so
19:07
<franksalim>
MIME sockets???
20:34
<Workshiva>
I note that wikipedia deleted the article about DeathStation 9000
20:37
<annevk5>
deletionists strike again?
20:43
<jgraham>
I want to write an article about deletionists and see how long it lasts
20:44
<annevk5>
I'm still not over what they did to the Pokeman pages
20:44
<annevk5>
of course I only started caring after I learned about what happened (though I had seen the original ones as well for unclear reasons)
20:45
<Workshiva>
Gotta protect those bytes, so people don't waste them
20:50
<hsivonen>
wikipedia deletionism is sad and annoying
20:51
<hsivonen>
particularly when you want to know what something is or what someone is supposed to be famous for and you see that there has been an article but is has been deleted as not notable enough
20:52
jgraham
wonders how articles about TV shows "are more likely to result in articles that are OR"
20:53
<jgraham>
How can watching a TV show and summarising the plot count as original research?
20:53
<jgraham>
The source... is the TV show
20:53
<Workshiva>
First party sources aren't good
20:54
<jgraham>
That's clearly insane
20:55
<jgraham>
If there is a track listing for a CD (common on wikipedia) I don't also expect there to be a link to an independent source listing the tracks on the CD
20:55
<AryehGregor>
It's completely insane, but there you have it.
20:56
<GPHemsley>
Is there a way to excluded a form element (e.g. <input type="submit">) from being submitted in a GET request?
20:56
<GPHemsley>
-d
20:56
<AryehGregor>
Primary sources that have been reliably published may be used in Wikipedia, but only with care, because it is easy to misuse them. Any interpretation of primary source material requires a reliable secondary source for that interpretation. A primary source may only be used on Wikipedia to make straightforward, descriptive statements that any educated person, with access to the source but without specialist knowledge, will be able to verify are
20:56
<AryehGregor>
supported by the source. For example, an article about a novel may cite passages to describe the plot, but any interpretation needs a secondary source. Do not make analytic, synthetic, interpretive, explanatory, or evaluative claims about material found in a primary source. Do not base articles entirely on primary sources. Do not add unsourced material from your personal experience, because that would make Wikipedia a primary source of that
20:56
<AryehGregor>
material.
20:56
<AryehGregor>
Blech, too long.
20:57
GPHemsley
tries to avoid the Wikipedia discussion
20:57
<AryehGregor>
GPHemsley, why do you want to not submit it?
20:58
<GPHemsley>
because, in essence, the form is intended to just add a sort variable to the URL, it doesn't need the submit variable too
20:58
<GPHemsley>
(whether that is a Bad Thing™, I don't know)
21:01
<AryehGregor>
You can't just ignore the submit variable?
21:03
<GPHemsley>
I want pretty URLs :)
21:03
<Workshiva>
GPHemsley: Don't give it a name?
21:04
<GPHemsley>
oh, that's easy :P
21:04
<GPHemsley>
thank you
21:05
<hsivonen>
GPHemsley: see the script on validator.nu
21:05
<GPHemsley>
hsivonen: oh, is that bad?
21:06
<hsivonen>
jgraham: OTOH, wikipedia seems to be fine with using XHTML2 WG's publication themselves as sources when writing about the XHTML2 WG's publications...
21:06
<AryehGregor>
hsivonen, it looks like you can directly repeat what primary sources say, you just can't summarize or interpret them.
21:07
<hsivonen>
GPHemsley: the script on validator.nu does tricks to make GET URLs shorter
21:07
<GPHemsley>
oh, well, I don't want JS tricks :)
21:08
<hsivonen>
AryehGregor: I mean even the propaganda parts of XHTML2 WG's publications
21:08
<GPHemsley>
hsivonen: Agh, why is the view-source of validator.nu afraid of line endings? ;_;
21:09
<AryehGregor>
Well, it's a secondary source on that, I guess. It shouldn't be taken as fact if it's disputed, it should be cited as the XHTMLWG's opinion.
21:11
<hsivonen>
GPHemsley: it's machine to machine communications :-)
21:12
<GPHemsley>
yeah, well, this machine doesn't like horizontal scrolling ;)
21:57
<jgraham>
This is a public service announcement (with guitars): chocolate from Trinidad is *so* nice
21:57
<jgraham>
At least the stuff I have right now is
21:58
<jgraham>
And I think the single origin Valrhona I particularly liked was from there too
21:58
<jgraham>
But I am not sure
22:00
<jgraham>
Also, it is really annoying how hard it is to do the simplest things without running into browser bugs
22:01
<jgraham>
It is doubly annoying when you were the one who should have found the bug
22:01
<jgraham>
One day someone will write a musical about the tragic life of a browser QA
22:02
<jgraham>
(I would have said opera, but that could have been mistaken for a pun)
22:02
<gsnedders>
jgraham: I find you notice minor issues far less when not a browser QA
22:03
<boblet>
jgraham: chocolate announcement requires independent verification, please forward to…
22:03
<boblet>
completely agree re: bugs too
22:06
<boblet>
regarding the a element, it says “Contexts in which this element may be used: Where phrasing content is expected”. As a can be flow content, shouldn’t that be where flow or phrasing content is expected?
22:08
<jgraham>
Oh
22:09
<jgraham>
I just discovered a site selling catering-sized bags of chocolate
22:09
<jgraham>
That is *such* a bad thing to know about
22:09
<gsnedders>
Will all end well?
22:09
<jgraham>
At 60GBP a bag, it would end in poverty at least
22:10
<jgraham>
(3kg)
22:10
<gsnedders>
You'll need a lot more rasberies to make up for that…
22:10
<jgraham>
Yeah
22:10
jgraham
closes that tab quickly
22:11
<boblet>
you need the equivalent of a pet food dispenser to limit consumption. it’d work then
22:15
<boblet>
guess I’ll file a bug, although I can’t believe no one has noticed it until now, which makes me wonder if I’m missing something re: flow/phrasing content :/
22:15
<gsnedders>
Well, a lot of browser people only care about UA requirements
22:55
<jgraham>
Hmm, i think I just managed to DOS myself
22:55
<jgraham>
s/DOS/DoS/
22:56
<jgraham>
I didn't mistakenly install a legacy operating system, after all
22:57
<boblet>
jgraham: that would have been far worse
22:59
<jgraham>
boblet: Well if this doesn't finish soon my laptop will be so hot it will set fire to my house
22:59
<jgraham>
and I will probably die in the flames
22:59
<jgraham>
So I'm not sure I agree with your assessment
23:00
<boblet>
jgraham: you can put some water on it to cool it down
23:00
jgraham
notes never to let boblet join the fire brigade
23:01
<boblet>
jgraham: good point, ice would be better huh
23:01
<annevk>
you guys missed http://www.w3.org/TR/2010/REC-xhtml-modularization-20100729/ ?
23:02
<jgraham>
annevk: I think "missed" is not quite right
23:02
<jgraham>
Maybe "filtered out as a pointless waste of bits"
23:03
<jgraham>
Although I did click on the link
23:03
<jgraham>
So it may open sometime in the distant future if the browser ever starts responding again
23:03
<annevk>
glad to waste your time ;p
23:04
<annevk>
maybe I'll get some sleep
23:04
<annevk>
got to clean up some stuff tomorrow morning and get to Maastricht in time for some hopefully constructive discussion
23:05
<boblet>
good to see they’re making progress on that whole XHTML thing
23:08
jgraham
wonders when Hixie is looking at WebSockets
23:09
<annevk>
if anyone wants to have me say anything on WebSockets at the Hybi meeting tomorrow just email and I'll try tor relay
23:10
<TabAtkins>
"stop complicating things, yor morons"?
23:10
<franksalim>
"sockets don't send messages"
23:10
<annevk>
heh, I'm gonna try to work out our different points of view; not sure if "morons" will be effective at that :)
23:11
<jgraham>
franksalim: Hmm? Websockets do at the moment
23:11
<TabAtkins>
"stop complicating things, you wonderful people"?
23:11
<franksalim>
ok, "sockets don't send MIME messages"
23:11
<annevk>
jgraham, I think what he means is that they don't need a media type and all
23:11
<annevk>
right
23:11
<jgraham>
Yeah
23:11
<franksalim>
some of the proposals are so far away from "TCP for the Web" it hurts
23:12
<Hixie>
jgraham: sup?
23:12
<annevk>
I think my main input is gonna be 1) settle on handshake 2) settle on framing 3) keep it very simple; and generally try to explain where we are coming from
23:12
<jgraham>
Hixie: sup??
23:13
<jgraham>
(I have no idea what "sup?" means)
23:13
<TabAtkins>
"what's up"
23:13
<jgraham>
Is this some Google thing
23:13
<annevk>
no
23:14
<TabAtkins>
No, it's an american thing
23:14
<jgraham>
Are you all developing your own langauge
23:14
<annevk>
sup? is pretty normal
23:14
<annevk>
you should watch some more series or movies
23:14
<TabAtkins>
jgraham isn't down with the kid's slang these days
23:14
<Hixie>
jgraham: what do i need to look at?
23:14
<annevk>
lol
23:14
<annevk>
(at Hixie)
23:15
<jgraham>
Hixie: Well there has been a lot of discussion on the list
23:15
<jgraham>
Some more insane than others
23:16
<jgraham>
I think it would be nice to get some input from you, at least about the framing stuff, soon given the "4 week" timetable
23:16
<jgraham>
(much of the other stuff is irrelevant and can be ignored)
23:16
<Hixie>
k
23:16
<jgraham>
(at least in the short term)
23:16
<Hixie>
i looked earlier and didn't see anything new on the framing stuff
23:16
<Hixie>
but i'll look at the newer e-mails
23:16
<annevk>
the chunking stuff seems somewhat compelling
23:17
<Hixie>
chunking is pointless since the client can't do anything until it's received the whole message anyway, as far as i can tell
23:17
<annevk>
according to Yngve something like that might even get more efficient than varied length-encoding
23:17
<jgraham>
Basically at this point I think the IETF people will conclude that they have consensus around length-delimited frames
23:17
<jgraham>
s/delimited/marked/
23:17
<annevk>
chunking is for the server to not have to know the length of a very large message
23:17
<franksalim>
i think lengh prefixing is superior to a terminator, myself
23:18
<Hixie>
length prefixing when the data is UTF-8 is just asking for subtle bugs
23:18
<franksalim>
annevk, why not send fixed sized messages at the socket layer and implement chunking in your protocol?
23:18
<Hixie>
that's a non-starter
23:18
<franksalim>
plenty of protocols over tcp have chunking
23:18
<Hixie>
annevk: for text you don't need to know the length at all
23:18
<franksalim>
Hixie, i think the terminator approach asks for different, complimentary bugs
23:19
<annevk>
franksalim, yeah, I suppose you can implement chunking yourself
23:19
<Hixie>
franksalim: like what?
23:19
<franksalim>
for instance if server code allows non string data to be sent as strings accidentally
23:19
<franksalim>
i have already seen websocket libraries that make that mistake
23:19
<franksalim>
written by professionals no less
23:19
<annevk>
Hixie, yeah, this is for non-text; though some people seem to dislike the sentinel stuff too, but I'm not too convinced myself
23:19
<franksalim>
that accept byte slices and send as UTF-8
23:19
<Hixie>
franksalim: well if people are sending non-UTF-8 content, it doesn't much matter if they're sending lengths or terminators, the data is bogus anyway
23:20
<Hixie>
annevk: for non-text, we can do chunking, sure. That's a non-issue now since you can't send non-text.
23:20
<jgraham>
One argument for length markers was that it helps pre-allocating buffers
23:20
<franksalim>
why would we do chunking for one type and not another?
23:21
<Hixie>
so that people who don't need it don't have to do it
23:21
<franksalim>
also, i agree with folks on the list that it is best to spec binary now even though javascript will not be able to send/recv it
23:21
<jgraham>
Which seems to be good for avoiding typical buffer overflow bugs
23:21
<Hixie>
why?
23:21
<Hixie>
(^to franksalim)
23:21
<franksalim>
Hixie, i don't see why the api and the protocol need to be exactly in lock step
23:21
<jgraham>
Yay! Firefox unfroze!
23:21
<annevk>
binary is more or less defined already
23:21
<franksalim>
Hixie, i think we can _finish_ the protocol before TypedArray
23:22
<annevk>
it's %80 iirc
23:22
<annevk>
it's just invalid for now and not exposed
23:22
<Hixie>
franksalim: without knowing what the API is we can't design the protocol
23:22
<franksalim>
annevk, with the provision that type bytes will be defined later, right?
23:22
<franksalim>
Hixie, i don't think that is true
23:22
<Hixie>
jgraham: i don't see how you could possibly have a buffer overflow if you're not relying on someone telling you the length of the data
23:22
<annevk>
type bytes?
23:22
<jgraham>
It ran for 1763 seconds and then popped up the slow script dialog...
23:22
<annevk>
you mean like media streams?
23:22
<franksalim>
Hixie, i have put different APIs over the same protocol. I have also put the same API over different protocols
23:23
<Hixie>
jgraham: in fact, you're far more likely to have a buffer overflow with a length byte, since it's more likely that you'll allocate a buffer and then mistakenly inspect the data to read it in
23:23
<Hixie>
franksalim: i can't design a protocol without knowing the api.
23:24
<Hixie>
you can always jam things together and make them work, but that's not a good situation to be in.
23:24
<franksalim>
Hixie, what can i say to that?
23:24
<annevk>
while I'm gonna read and sleep some, keep this up; it's nice input :)
23:24
<franksalim>
Hixie, I can?
23:24
<jgraham>
Hixie: Anyway, it would be good if you have this discussion on the list
23:24
<Hixie>
we've had this discussion on the list
23:24
<Hixie>
multiple times over the last year or two
23:24
<jgraham>
Personally I am happy with either type of framing for text
23:25
<Hixie>
i don't see any point repeating what i've already said
23:25
<jgraham>
Hixie: I am pretty much 100% convinced that people will not be happy with that
23:26
<Hixie>
i'm not trying to make people happy, i'm trying to design a protocol
23:26
<jgraham>
Hixie: I don't think they see it like that either
23:26
<Hixie>
that's their problem :-)
23:26
<jgraham>
I think they think that the group is trying to design a protocol
23:28
<franksalim>
this can of worms would have been avoided completely by having the protocol be byte streams (like tcp) instead of discrete messages
23:28
<franksalim>
with byte streams it is clear you can do whatever you want and it is up to the next protocol up the stack to define structure, types, chunking, muxing
23:29
<Hixie>
byte streams would have to be exposed to JS, and I'm not at all convinced that your typical web dev can write code to handle byte streams well
23:30
<franksalim>
a typical web dev can't use opengl, either
23:30
<boblet>
are there any UTF-8 characters apart from & and > that need escaping in HTML5?
23:30
<jgraham>
Yeah, I think byte streams would be bad
23:30
<franksalim>
messaging protocols sitting on top of byte streams are equivalent to scene graphs on webgl
23:30
<TabAtkins>
boblet - you mean <, right?
23:31
<Hixie>
franksalim: i wasn't especially convinced that we should expose webgl either
23:31
<boblet>
TabAtkins: doh, yep
23:31
<TabAtkins>
Then in most circumstances, yes. In some circumstances you'll want to escape ',", and >.
23:32
<franksalim>
Hixie, if one messaging protocol 'wins,' we can bake it in
23:32
<boblet>
TabAtkins: aah good point. you mean eg in alt text yeah?
23:33
<TabAtkins>
Yeah.
23:33
<franksalim>
starting with a stripped down or half baked messaging protocol is the worst idea, imho
23:33
<TabAtkins>
If you're escaping attributes, escape ' and ". If you're escaping unquoted attributes, also escape > (as it can early-close the element in some circumstances).
23:34
<jgraham>
Anyway, bedtime
23:34
<jgraham>
Goodnight
23:34
<boblet>
jgraham: happy chocolate dreams
23:34
<jgraham>
:)
23:35
<franksalim>
Hixie, I should say that I am much happier with the frames we currently have than anything heavier weight that has been proposed
23:36
<franksalim>
because lightweight frames allow me to ignore framing and run a real protocol over a stream
23:43
<franksalim>
if your application messages and your wire messages are one and the same, then all the problems like chunking and addressing that have been raised are valid. if you treat websocket like tcp, you handle those concerns in your protocol of choice
23:43
<AryehGregor>
Why do all of these tests fail on Chrome? It's puzzling me. I don't seem to be able to easily make a reduced test case. http://aryeh.name/tmp/html5tests.html
23:43
<AryehGregor>
(not all, 20%, but whatever)
23:43
<AryehGregor>
(I'm idly writing some tests for basic easy-to-test-from-JS stuff, dunno if anyone has written any such so far)